Journal paper software archive for Zenodo

A release that contains an image of the code on github, 01/13/2022, for journal paper publishing

v1.2.5 doodler paper

Version of Doodler used for the Doodler paper (forthcoming)

gifs for doodler README

7/27/21

example results from the Doodler program

example-results

7/27/21

Sample data for the Doodler program

Sample data for the Doodler program: 15 jpeg images of shoreline environments

v1.2.1

See https://dbuscombe-usgs.github.io/dash_doodler/blog/2021/05/16/blog-post for details

Version of the code to be used for the forthcoming Doodler manuscript

Doodler v1.1.1, March 17 2021

First release to enforce the use of relative pixel location for both RF and CRF inference. Results in improved task-specific estimates and predictions

Doodler v1.1.0, March 17 2021

release-v1.1.0

v1.0.0

This is the first release of dash-doodler, to coincide with the first version that appears to work well for multiple data sets and class sets, using both versions (appCRF.py and appRF.py).

This serves to snapshot this version of the program ahead of major planned changes, involving a database backend, user ID, and web serving.

Subsequent updates will follow major.minor.build convention

the CRF implementation is now faster and better. The RF implementation works well on all tested data, too. the defaults for CRF parameters have changed. Also, the image data representation it is using has changed, and the segmentation now uses nearest-neighbour rather than linear interpolation, which makes more sense for these label images with discrete classes. New sliders are now available to control the data density (increase the downsample factor for larger images). Note the new behavior of the median filter slider – median filtering still occurs if the value > 1, but won’t automatically redo the segmentation when its value is changed (to do that, you should recheck the compute/show segmentation box). The program now automatically selects a new colormap if you have more than 10 classes.